Content uploaded by Shannon M Doyle
Author content
All content in this area was uploaded by Shannon M Doyle
Content may be subject to copyright.
Coupling ATP utilization to protein remodeling
by ClpB, a hexameric AAAⴙprotein
Joel R. Hoskins
1
, Shannon M. Doyle
1
, and Sue Wickner
2
Laboratory of Molecular Biology, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892
Contributed by Sue Wickner, October 22, 2009 (sent for review August 14, 2009)
ClpB and Hsp104 are members of the AAAⴙ(ATPases associated
with various cellular activities) family of proteins and are molecular
machines involved in thermotolerance. They are hexameric pro-
teins containing 12 ATP binding sites with two sites per protomer.
ClpB and Hsp104 possess some innate protein remodeling activi-
ties; however, they require the collaboration of the DnaK/Hsp70
chaperone system to disaggregate and reactivate insoluble aggre-
gated proteins. We investigated the mechanism by which ClpB
couples ATP utilization to protein remodeling with and without the
DnaK system. When wild-type ClpB, which is unable to remodel
proteins alone in the presence of ATP, was mixed with a ClpB
mutant that is unable to hydrolyze ATP, the heterohexamers
surprisingly gained protein remodeling activity. Optimal protein
remodeling by the heterohexamers in the absence of the DnaK
system required approximately three active and three inactive
protomers. In addition, the location of the active and inactive ATP
binding sites in the hexamer was not important. The results
suggest that in the absence of the DnaK system, ClpB acts by a
probabilistic mechanism. However, when we measured protein
disaggregation by ClpB heterohexamers in conjunction with the
DnaK system, incorporation of a single inactive ClpB subunit
blocked activity, supporting a sequential mechanism of ATP utili-
zation. Taken together, the results suggest that the mechanism of
ATP utilization by ClpB is adaptable and can vary depending on the
specific substrate and the presence of the DnaK system.
DnaJ 兩DnaK 兩GrpE 兩protein disaggregation
Bacterial ClpB and yeast Hsp104 are ATP-dependent protein
remodeling machines that function to disaggregate protein
aggregates and reactivate proteins after extreme stress condi-
tions (1–3). In the cell, ClpB acts in conjunction with the DnaK
chaperone system and Hsp104 acts with the Hsp70 chaperone
system (4, 5). DnaK and Hsp70 are members of another large,
ubiquitous family of ATP-dependent molecular chaperones that
mediate protein reactivation and remodeling in concert with two
cochaperones, DnaJ and GrpE in prokaryotes and Hsp40 and
NEF in eukaryotes (6). Alone, neither the DnaK/Hsp70 chap-
erone system nor ClpB/Hsp104 has the ability to reactivate large
insoluble aggregates.
ClpB/Hsp104 exists as a hexameric ring with an axial channel
(7–10). Each protomer contains two AAA⫹(ATPases associ-
ated with various cellular activities) nucleotide-binding domains
separated by a hinge region and preceded by an N-terminal
domain (1, 7). The two AAA⫹domains contain characteristic
motifs, including Walker A and B and sensor-1 and -2 motifs, as
well as an arginine finger (11, 12). Situated in the first AAA⫹
domain is a long coiled-coil region, referred to as the middle
domain, which is unique to ClpB, Hsp104, and their homologs.
In vitro ClpB/Hsp104 solubilizes and reactivates protein ag-
gregates in ATP-dependent reactions in collaboration with the
DnaK/Hsp70 chaperone system (1–3). Although the roles of the
two chaperone systems in disaggregation are not fully under-
stood, it is likely that ClpB/Hsp104 is the primary protein
disaggregating machine, and DnaK/Hsp70 facilitates the inter-
action of ClpB/Hsp104 with aggregates (1–3). However, DnaK/
Hsp70 may have additional functions.
Evidence suggests that ClpB/Hsp104 acts by forcibly extracting
polypeptides from aggregates and translocating the unfolded
regions of polypeptides through its axial channel (1–3). This
substrate translocation mechanism has been established for
ClpA and ClpX, two Clp ATPases that interact with a proteolytic
component, ClpP (13). Substrates are unfolded and threaded
through the central channels of ClpA and ClpX directly into the
chamber of ClpP where degradation occurs. Work from Bukau
and colleagues, in which ClpB and Hsp104 were engineered to
contain the ClpP interaction loop from ClpA, supports a similar
mechanism for ClpB and Hsp104 (14, 15).
The demonstration that ClpB and Hsp104 have an innate
ability to unfold and activate proteins in the absence of the
DnaK/Hsp70 chaperone system provides additional support for
an unfolding and translocation mechanism (16). However, ATP
alone is ineffective in promoting these protein remodeling
activities. Instead, mixtures of ATP and ATP
␥
S, a slowly hy-
drolyzed ATP analog, are required for remodeling, suggesting
that with the mixture of nucleotides substrates can be held (a
function requiring ATP binding and supported by ATP
␥
S) and
unfolded and translocated (functions requiring ATP hydrolysis).
To understand how ClpB couples ATP binding and hydrolysis
to protein remodeling, we investigated the contribution of the 12
ATP binding sites to the overall chaperone activity of ClpB both
alone and in the presence of the DnaK chaperone system.
Results
ClpB Hexamers with a Balance of Active and Inactive Nucleotide
Binding Sites Are Required for Optimal Protein Remodeling in the
Absence of the DnaK System. The 12 ATP binding sites of hex-
americ ClpB are arranged in two rings of six, with each protomer
providing one nucleotide-binding site to each ring, referred to
here as Ring-1 and Ring-2 (Fig. 1A). Ring-1 of the hexamer is
comprised of the N-terminal ATP binding sites, and Ring-2 is
comprised of the C-terminal ATP-binding sites.
Previous observations demonstrated that ATP hydrolysis at
⬍12 ATP binding sites of the ClpB hexamer is required to elicit
the innate protein remodeling activity of ClpB, and AT P hy-
drolysis at all 12 sites prohibits activity (16). In the work
presented here, we used a green fluorescent protein (GFP)
fusion protein containing a C-terminal 15 aa peptide, GFP-15,
as a substrate for remodeling. We measured protein unfolding by
monitoring the decrease in fluorescence with time in the pres-
ence of a mutant GroEL (17) that binds unfolded proteins but
does not release them. ClpB wild-type, ClpB
(wt)
, was unable to
unfold GFP-15 in the presence of ATP (Fig. 1B). However, when
mixtures of ATP and ATP
␥
S were used in a 1:1 ratio, there was
a large decrease in GFP fluorescence (Fig. 1C). Thus, if some
Author contributions: J.R.H., S.M.D., and S.W. designed research; J.R.H. and S.M.D. per-
formed research; J.R.H. and S.M.D. contributed new reagents/analytic tools; J.R.H., S.M.D.,
and S.W. analyzed data; and J.R.H., S.M.D., and S.W. wrote the paper.
The authors declare no conflict of interest.
1J.R.H. and S.M.D. contributed equally to this work.
2To whom correspondence should be addressed. E-mail: wickners@mail.nih.gov.
This article contains supporting information online at www.pnas.org/cgi/content/full/
0911937106/DCSupplemental.
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0911937106 PNAS
兩
December 29, 2009
兩
vol. 106
兩
no. 52
兩
22233–22238
BIOCHEMISTRY
ATP binding sites hydrolyze ATP slowly, ClpB
(wt)
activity is
elicited.
Single Walker B mutants (Fig. 1 A), with either an E279A
substitution in Ring-1, ClpB
(B1)
, or an E678A substitution in
Ring-2, ClpB
(B2)
, unfolded GFP-15 with ATP alone and were
inhibited by ATP
␥
S, supporting the conclusion that hydrolysis by
⬍12 ATP binding sites elicits activity (Fig. 1 Band C). These
mutants contain six wild-type nucleotide-binding sites per hex-
amer and six mutant sites that are able to bind but not hydrolyze
ATP (18). A double Walker B mutant (Fig. 1 A) with both E279A
and E678A substitutions, ClpB
(B1,B2)
, was unable to catalyze
unfolding of GFP-15 with ATP or mixtures of ATP and ATP
␥
S,
since all of its nucleotide-binding sites are defective in ATP
hydrolysis (19) (Fig. 1 Band C). Therefore, ClpB can perform
protein unfolding when there is only one active nucleotide-
binding site per protomer, and the active nucleotide-binding sites
are all in Ring-1 or Ring-2 of the hexamer. These results also
show that hydrolysis by the two rings does not need to be coupled
for protein remodeling by ClpB alone.
To investigate whether the intrinsic remodeling activity of
ClpB demands that all six of the active sites reside in Ring-1 or
Ring-2 of the hexamer, or if the active sites can be distributed
between the two rings, we performed subunit mixing experi-
ments. Recent studies have shown that the ClpB hexamer is a
dynamic complex with subunits reshuffling on a time scale of a
minute (20, 21). Thus, in subunit mixing experiments, hetero-
hexamers are expected to be represented by a binomial distri-
bution that varies as a function of the molar ratio of each subunit
present, assuming subunits have an equal ability to be incorpo-
rated into hexamers (Fig. 1D). We mixed the double Walker B
mutant, ClpB
(B1,B2)
, with ClpB
(wt)
(shown schematically in Fig.
1E) and measured GFP-15 unfolding in the presence of ATP as
the sole nucleotide. Surprisingly, we saw that the heterohexam-
ers were active, although each protein alone was inactive in
GFP-15 unfolding (Fig. 1F). By var ying the ratio of ClpB
(wt)
and
ClpB
(B1,B2)
while holding the total ClpB concentration constant,
we observed maximal activity when there were approx imately
three active sites in each ring of the hexamer [50% ClpB
(wt)
and
50% ClpB
(B1,B2)
] (Fig. 1 Fand G). Therefore, ClpB is able to
catalyze protein remodeling when there are both active and
inactive sites in Ring-1 and Ring-2 of the hexamer and the
subunits contain either two active or two inactive nucleotide-
binding sites.
We extended these observations with ClpB
(B1,B2)
and ClpB
(wt)
heterohexamers by testing two other protein remodeling reac-
tions. In a disaggregation assay, mixtures of ClpB
(B1,B2)
and
ClpB
(wt)
were able to reactivate heat-inactivated GFP in the
presence of ATP alone (Fig. 2A). Maximal reactivation was
ClpB(wt)
(side view of hexamer) ClpB(B1)
Ring-1
Ring-2
ClpB(B2)
A
BC
Sites active for ATP
hydrolysis (red)
2 nucleotide-binding
sites in a protomer (connection
shown by white oval)
Sites able to bind but not
hydrolyze ATP (blue)
1:1 mixture of
ClpB(wt) and ClpB(B1,B2)
ClpB(B1)
ClpB(B2)
ClpB(wt)
ClpB(B1,B2)
ClpB(B1,B2)
ClpB(wt):ClpB(B1,B2)
1.0
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0
ClpB(B1)
ClpB(B2)
ClpB(wt)
ClpB(B1,B2)
0 10 20 30 40 0 10 20 30 40
Time (min) Time (min)
E
Time (min)
0 10 20 30 40 50 60
Fluorescence intensity (AU)
Fluorescence intensity (AU)
Fluorescence intensity (AU)
ATP and ATP S
ATP
FATP
0:6
1:5
6:0
5:1
4:2
3:3
2:4
G
Mutant (%)
0 20 40 60 80 100
Population (fraction)
1.0
0.8
0.6
0.4
0.2
0.0
D
GFP-15 unfolded (min-1)
0 20 40 60 80 1 00
0.03
0.02
0.01
0.00
ClpB(wt) (%)
ClpB(B1,B2) (%)
100 80 60 40 20 0
Fig. 1. Optimal protein unfolding by ClpB requires that ClpB hexamers have both active and inactive ATP hydrolytic sites. (A) Diagram illustrating the location
of the 12 ATP binding sites in hexamers of ClpB(wt), single Walker B mutants in the first, ClpB(B1), or second, ClpB(B2), nucleotide-binding domain and a double
Walker B mutant, ClpB(B1,B2). The 12 sites are superimposed on a model of a ClpB hexamer generated from the crystal structure of a ClpB monomer (39). (Band
C) Protein unfolding of GFP-15 by ClpB(wt), ClpB(B1), ClpB(B2), and ClpB(B1,B2) in the presence of ATP (B) or ATP and ATP
␥
S in a 1:1 ratio (C) as described in the Methods.
(D) Mathematical model, generated as described in ref. 20, depicting the theoretical populations of wild-type and mutant ClpB heterohexamers that contain
no (black), one (aqua), two (red), three (green), four (purple), five (orange), or six (blue) mutant subunits as a function of percent mutant. (E) Diagram showing
representative locations of wild-type (red) and Walker B mutant ATP binding sites (blue) for a heterohexamer of ClpB(wt) and ClpB(B1,B2) in a 1:1 ratio. (F) GFP-15
unfolding in the presence of ATP using mixtures of ClpB(wt) and ClpB(B1,B2) in various ratios. (G) The rate of GFP-15 unfolding by mixtures of ClpB(wt) and ClpB(B1,B2)
plotted as a function of percent ClpB(B1,B2) in the mixture. In B,C, and F, representative experiments of three replicates are shown. In G, data are means ⫾SEM
(n⫽3). Some error bars are covered by the plot symbols.
22234
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0911937106 Hoskins et al.
observed with a 1:1 ratio of the two ClpB proteins. Mixtures of
ClpB
(B1,B2)
and ClpB
(wt)
were also tested in the RepA activation
assay in which inactive RepA dimers are converted to monomers
that bind DNA with high affinity (22). In this assay as well,
maximal activation was seen with a 1:1 mixture of the two
proteins (Fig. 2A). Thus, in additional protein remodeling
reactions, heterohexamers of ClpB
(B1,B2)
and ClpB
(wt)
are func-
tional, while neither protein separately exhibits significant
activity.
When ATP hydrolysis was measured, approximately 2-fold
higher steady-state hydrolysis was observed with a 1:1 mixture of
ClpB
(wt)
and ClpB
(B1,B2)
compared to ClpB
(wt)
alone (Fig. 2B)
(19). This suggests that the activity of wild-type subunits within
the heterohexamer is stimulated. For example, at a 1:1 ratio of
ClpB
(wt)
to ClpB
(B1,B2)
, the ATPase activity of the wild-type
subunits must increase approximately 4-fold to account for the
observed increase in ATP hydrolysis. For comparison, when we
measured ATP hydrolysis by mixtures of ClpB
(wt)
and ClpB
(B1,B2)
in the presence of ATP and ATP
␥
S, hydrolysis decreased
linearly as the percent of ClpB
(B1,B2)
in the mixture was increased
(Fig. 2B). AT P hydrolysis by ClpB
(wt)
alone was approximately
2-fold higher with a 1:1 mixture of ATP and ATP
␥
S than with
ATP alone (Fig. 2 B) (16). Thus, conditions optimal for protein
remodeling are also optimal for ATP hydrolysis.
Our results with the single Walker B mutants and heterohex-
amers of ClpB
(wt)
and ClpB
(B1,B2)
suggest that for remodeling
activity by ClpB alone: (i) protomers need not contain a pair of
active sites as long as there are a total of six active sites in either
Ring-1 or Ring-2 of the hexamer, and (ii) there need not be six
functional sites in a single ring as long as there are approximately
six active sites in the hexamer. Together these results indicate
that the innate remodeling activity of ClpB might simply require
a total of approximately six active AT P hydrolytic sites, inde-
pendent of the location of the sites within the subunit or within
the hexamer.
To test this possibility, we measured protein unfolding by
heterohexamers of the two single Walker B mutants, ClpB
(B1)
and ClpB
(B2)
(Fig. 3A). When ClpB
(B1)
and ClpB
(B2)
were mixed
in various ratios while keeping the total concentration of ClpB
constant, there was no significant gain or loss of function (Fig.
3B). Interpretation of these results depends upon ClpB
(B1)
and
ClpB
(B2)
forming heterohexamers, which is likely, since each
single Walker B mutant is capable of forming heterohexamers
with both ClpB
(wt)
and ClpB
(B1,B2)
(shown in Fig. 4 Band Dand
Fig. 5B). The results suggest that ClpB is active in protein
remodeling when the hexamer contains six active AT P hydrolytic
sites irrespective of the position of the active sites within the
hexamer.
Mechanism of ATP Utilization by ClpB in the Absence of the DnaK
System. Our observations that heterohexamers of ClpB
(wt)
and
ClpB
(B1,B2)
perform remodeling activities demonstrate that pro-
tein unfolding and remodeling can be carried out with approx-
imately three active protomers per hexamer (Figs. 1 and 2).
These results rule out a concerted mechanism of ATP utilization
within a ring, where all six active sites of a ring simultaneously
bind ATP, hydrolyze ATP, and release ADP. They also rule out
a strictly sequential mechanism, where activity depends upon an
endless cycle of ATP utilization around the ring. Instead, the
observations point to either a probabilistic mechanism, involving
a random order of mutually independent hydrolysis events or a
semisequential mechanism, in which the sequence and timing of
ATP hydrolysis operates in a sequential manner that proceeds
around the ring with hydrolysis pausing occasionally and then
resuming at another site in the ring.
To further explore the mechanism of ATP utilization by ClpB,
we varied the number of active sites in one ring while maintaining
six inactive sites in the other ring. We first decreased the number
of active sites in Ring-2 by increasing the percentage of
ClpB
(B1,B2)
in mixtures with ClpB
(B1)
(Fig. 4A). We observed that
incorporation of one or two inactive sites in Ring-2 had little
effect on hexamer activity, while incorporation of four or five
inactive sites significantly inhibited activity (Fig. 4B). We next
changed the number of active sites in Ring-1 by varying the ratio
of ClpB
(B1,B2)
in mixtures with ClpB
(B2)
(Fig. 4C) and observed
similar results (Fig. 4D). Taken together, these experiments show
that ClpB is able to function in protein remodeling with one
ATP hydrolyzed (nmol)
100 80 60 40 20 0
0 20 40 60 80 100
B
A
GFP reactivation (% min-1)
0.06
0.04
0.02
0.00
3
2
1
0
RepA activation
(fmol oriP1 DNA bound)
100 80 60 40 20 0
0 20 40 6 0 80 100
ClpB(wt)
(%)
ClpB(B1,B2) (%)
ClpB(wt)
(%)
ClpB(B1,B2) (%)
25
20
15
10
5
0
ATP
ATP + ATP S
GFP
reactivation
RepA
activation
Fig. 2. ClpB hexamers containing approximately six active ATP hydrolytic
sites are required for optimal chaperone activity. (A) Reactivation of heat-
aggregated GFP and activation of RepA was measured as described in the
Methods using various ratios of ClpB(wt) to ClpB(B1,B2). GFP reactivation rates
(left axis) and DNA binding by RepA (right axis) were plotted as a function of
percent ClpB(B1,B2).(B) ATPase activity by mixtures of ClpB(wt) and ClpB(B1,B2) in
the presence of ATP or ATP and ATP
␥
S in a 1:1 ratio was measured as described
in the Methods. ATPase activity is plotted as a function of percent ClpB(B1,B2).
Data are means ⫾SEM (n⫽3). Some error bars are covered by plot symbols.
A
1:1 mixture of ClpB(B1)
and ClpB(B2)
ClpB(B2) (%)
ClpB(B1) (%)
100 80 60 40 20 0
0 20 40 60 80 100
GFP-15 unfolded (relative)
Sites able to
bind but not
hydrolyze ATP
Active
sites
B
1.0
0.8
0.6
0.4
0.2
0.0
Fig. 3. Protein unfolding by mixtures of ClpB(B1) and ClpB(B2).(A) Diagram
showing a heterohexamer containing ClpB(B1) and ClpB(B2) in a 1:1 ratio. (B)
GFP-15 unfolding by mixtures of ClpB(B1) and ClpB(B2) was measured as de-
scribed in the Methods. Rates are expressed as a fraction relative to ClpB(B2)
alone (0.060 ⫾0.002 min⫺1) and plotted as a function of percent ClpB(B1) in the
mixture. Data are means ⫾SEM (nⱖ3). Some error bars are covered by plot
symbols.
Hoskins et al. PNAS
兩
December 29, 2009
兩
vol. 106
兩
no. 52
兩
22235
BIOCHEMISTRY
inactive ring and approximately three or more active sites in the
other ring of the hexamer (Fig. 4 Band D). Thus, the results best
support a probabilistic or a semisequential mechanism of ATP
utilization by the ClpB rings. However, six active sites per
hexamer are optimal for protein remodeling in the absence of the
DnaK system (Figs. 1 Fand Gand 2–4).
Mechanism of ATP Utilization by ClpB When Disaggregating Sub-
strates That Require the DnaK System. To test whether the mech-
anism of ATP utilization by ClpB is the same when ClpB works
on aggregates with the DnaK system as when it works alone, we
monitored disaggregation by mixtures of ClpB
(wt)
and
ClpB
(B1,B2)
in conjunction with DnaK, DnaJ, and GrpE. We used
aggregated malate dehydrogenase (MDH) as a substrate, since
previous studies showed that MDH disaggregation by ClpB in
combination with the DnaK system was approximately 20-fold
greater than disaggregation by either chaperone alone (23, 24).
As the percent of ClpB
(B1,B2)
in mixtures with ClpB
(wt)
was
increased, we observed a rapid exponential decrease in MDH
disaggregation, compared to the linear decrease expected if a
probabilistic mechanism was used (Fig. 5A). The loss of activity
when the percentage of ClpB
(B1,B2)
is low indicates that incor-
poration of approximately one protomer with two hydrolytically
defective ATP binding sites blocks disaggregation of MDH. Our
results indicate a striking departure from the protein remodeling
and ATP-ase activities seen in the absence of the DnaK system
(Figs. 1 Fand Gand 2 Aand B) and from the ATPase data (Fig.
2B). They suggest that for disaggregation of protein aggregates
in collaboration with the DnaK system, cooperative interactions
between all six ClpB subunits are required. Similar inter-subunit
coupling during remodeling has been observed for Thermus
thermophilus ClpB using another substrate, aggregated
␣
-gluco-
sidase (20).
We next measured the effects of single nucleotide binding site
mutants in mixtures with ClpB
(wt)
. We used the two single
Walker B mutants discussed above, ClpB
(B1)
and ClpB
(B2)
.In
addition, we tested a ClpB Walker A mutant, ClpB
(A2)
, with a
K611T substitution in Ring-2 that renders it unable to bind ATP
in Ring-2.* All three mutants are defective in disaggregation of
MDH in combination with the DnaK system (23, 25) (Fig. 5 B
and C). As the percentage of ClpB
(B1)
, ClpB
(B2)
, or ClpB
(A2)
in
mixtures with ClpB
(wt)
was increased, disaggregation of MDH
decreased more rapidly than the linear decrease expected for a
probabilistic mechanism of action (Fig. 5 Band C). However,
incorporating approximately two inactive sites in the same ring
inhibited disaggregation less than incorporating approximately
one protomer with two inactive sites (Fig. 5 Aand B). These
observations suggest that for ClpB to work on substrates that
require collaboration with the DnaK system, both inter- and
intra-ring communication between ClpB subunits is important.
Together the data support a sequential or semisequential mech-
anism of ATP utilization within each ring and within the
hexamer. Moreover, they imply that the mechanism of ATP
utilization by ClpB can vary depending on the substrate and the
requirement for the DnaK system.
To address the question of whether the specific aggregate
influences ATP utilization requirements, we compared disag-
gregation of aggregated MDH to that of heat-aggregated GFP-
38, a GFP fusion protein containing a C-terminal 38 aa peptide.
Aggregated GFP-38, like MDH, required the combination of
ClpB
(wt)
and the DnaK system for reactivation, as measured by
the regain of GFP fluorescence with time (Fig. 6A). ClpB
(B1)
and
ClpB
(B2)
were each able to reactivate GFP-38 at a low rate in the
presence of the DnaK system, although neither mutant was able
to detectably disaggregate MDH (Figs. 5Band 6 A). These results
suggest that with some aggregates, six active ATP binding sites
are sufficient to carry out limited disaggregation in conjunction
with the DnaK system.
To extend the comparison of the two aggregates, we measured
disaggregation of GFP-38 by mixtures of ClpB
(B1,B2)
and Clp-
B
(wt)
. Disaggregation activity decreased exponentially as the
ratio of ClpB
(B1,B2)
to ClpB
(wt)
in the mixture increased (Fig. 6B).
The data provide further support that cooperative interactions
between subunits are important and that ATP utilization by
ClpB is likely through a sequential or semisequential mechanism
when ClpB acts in protein disaggregation in conjunction with the
DnaK system.
In summary the results presented here suggest that the
mechanism of ATP utilization by the two rings of ClpB can
change dependent upon the specific substrate and the require-
ment for the DnaK chaperone system.
Discussion
The surprising result that the heterohexamer of ClpB
(wt)
and
ClpB
(B1,B2)
is active under conditions where neither homohex-
amer is active shows that modulation of the hydrolytic cycle of
ClpB elicits protein remodeling. Moreover, approximately six
active nucleotide-binding sites are required for maximal pro-
tein remodeling activity by ClpB alone (Figs. 1 and 2). These
observations are consistent with our previous findings that
ClpB and Hsp104 perform protein remodeling alone in the
presence of mixtures of ATP and ATP
␥
S (16). Together the
data suggest that remodeling activity is elicited when some
sites hydrolyze ATP and other sites bind ATP, but hydrolyze
*Unlike a Walker A mutant with a K611T substitution in Ring-2, a Walker A mutant with a
substitution in Ring-1, K212T, is unable to form hexamers (22, 27). For this reason,
ClpB(K212T) and a double Walker A mutant were not used.
Fig. 4. Protein unfolding by mixtures of ClpB(B1,B2) and either ClpB(B1) or
ClpB(B2).(Aand C) Diagrams showing heterohexamers of ClpB(B1,B2) with
ClpB(B1) (A) or ClpB(B2) (C) in a 1:1 ratio. (Band D) GFP-15 unfolding by mixtures
of ClpB(B1,B2) and ClpB(B1) (B) or ClpB(B2) (D) as described in the Methods.
Unfolding rates are expressed as a fraction relative to ClpB(B1) (0.035 ⫾0.001
min⫺1)(B) or ClpB(B2) (0.060 ⫾0.002 min⫺1)(D) alone and plotted as a function
of percent ClpB(B1,B2) in the mixture. Data are means ⫾SEM (nⱖ3). Some error
bars are hidden by plot symbols.
22236
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0911937106 Hoskins et al.
it slowly or not at all. One interpretation is that the sites that
bind but do not hydrolyze ATP help stabilize interactions
between ClpB and the substrate as well as interactions among
ClpB protomers in the hexamer, while the sites that hydrolyze
ATP are necessary for substrate unfolding and translocation.
The balance between the two types of sites influences the
efficiency of protein remodeling. However, modulating the
ATP hydrolytic cycle is not sufficient to elicit the full activity
of ClpB, since in vivo, mutants defective in AT P hydrolysis in
either ring are also defective for thermotolerance (18). One
possibility is that the substrate and the DnaK system regulate
ATP utilization, consistent with our previous results showing
that ATP hydrolysis is stimulated by the combined presence of
ClpB, the DnaK system, and aggregated substrate (23).
Interestingly, our results suggest that the number of hydro-
lytically active sites sufficient to perform remodeling activity
and the mechanism of ATP utilization can vary depending on
the substrate and the involvement of the DnaK system. When
ClpB acts in the absence of the DnaK system, the mechanism
of ATP utilization by the two rings is likely probabilistic,
although a semisequential mechanism is also possible. This is
based on experiments showing that approximately three active
protomers per hexamer are optimal for remodeling activity. In
addition, approximately three active ATP hydrolytic sites in a
ring are sufficient for protein remodeling when the other ring
contains six sites defective in ATP hydrolysis. In contrast,
when ClpB acts on aggregated substrates that require the
DnaK system, ClpB protomers must work together in a
cooperative fashion, supporting a sequential mechanism of
ATP utilization. However, the observations that heterohex-
amers can tolerate protomers with one active and one inactive
ATP binding site provide evidence for a semisequential mech-
anism of ATP utilization within the separate rings.
Probabilistic, semisequential, and sequential modes of ac-
tion have been proposed for other AA A⫹proteins (10, 26, 27).
For example, another hexameric Clp protein, ClpX, has been
shown to function in a probabilistic manner by Martin, Baker,
and Sauer (26). In contrast, Saibil, Lindquist, and colleagues
recently proposed that Hsp104, the yeast homolog of ClpB,
uses a sequential mechanism, based on electron microscopic
data showing asymmetry in the hexameric model (10). Crystal
structures showing asymmetric conformations have been ob-
served for several AAA⫹helicases, suggesting that those
proteins act by a sequential mechanism (27). Additionally,
subunit mixing studies provided evidence for a semisequential
mechanism of action by MCM and RuvB and a sequential
mechanism for T7gp4 (28–32).
While further biochemical and structural analyses of ClpB and
its interaction with the DnaK system are needed, the mechanistic
insights revealed by this study of ClpB may extend to Hsp104 and
other ClpB homologs.
Methods
Proteins and DNA. P1 RepA (22), GFP (33), GFP-15 (34), ClpB and ClpB mutants
(35), GroELtrap (17), DnaK (36), DnaJ (36), GrpE (36), and [3H]oriP1 DNA (22)
(4,475 cpm/fmol) were prepared as described. pET-GFP-38 was created by
inserting GAT and GAC codons before the multicloning site stop codon of
pET-GFP-X30-H6(37) by QuikChange (Stratagene) mutagenesis. GFP-38 was
purified as described for GFP-X30-H6(37). MDH (Roche) was labeled with 3Has
ABC
ClpB(wt) + ClpB(B1)
ClpB(wt) + ClpB(B2)
ClpB(wt) (%)
ClpB(B1,B2) (%)
0 20 40 60 80 100
ClpB(B1) or ClpB(B2) (%)
0 20 40 60 80 100
ClpB(A2) (%)
0 20 40 60 80 100
Soluble MDH (fraction)
Soluble MDH (fraction)
Soluble MDH (fraction)
1.0
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0
ClpB(wt) (%)
ClpB(wt) (%)
100 80 60 40 20 0 100 80 60 40 20 0
100 80 60 40 20 0
ClpB(wt) + ClpB(B1,B2)
ClpB(wt) + ClpB(A2)
Fig. 5. Disaggregation of MDH by mixtures of ClpB(wt) and ClpB mutants in conjunction with DnaK, DnaJ, and GrpE. (A–C) Recovery of soluble MDH from
heat-induced aggregates was measured as described in the Methods for mixtures of ClpB(wt) with ClpB(B1,B2) (A), ClpB(B1) (B), ClpB(B2) (B), or ClpB(A2) (C)inthe
presence of DnaK/DnaJ/GrpE. In A–C, the dashed gray line represents the linear decrease expected for a probabilistic mechanism where the activity of the hexamer
is proportional to the number of wild-type subunits. The fraction of soluble MDH recovered in reactions containing 100% ClpB(wt) was set equal to one. With
conditions used, ClpB(wt) solubilized 53% ⫾1% of the MDH. Data are means ⫾SEM (nⱖ3). Some error bars are covered by plot symbols.
No chaperone
K/J/E
ClpB(wt)
ClpB(wt) + K/J/E
ClpB(B1)
ClpB(B1) + K/J/E
ClpB(B2)
ClpB(B2) + K/J/E
A
B
GFP-38 reactivation (% min-1)
ClpB(B1,B2) (%)
GFP-38 reactivation
(% min-1)
0 1.0 2.0 3.0 4.0
3.0
2.0
1.0
0
0 20 4 0 60 80 100
ClpB(wt) (%)
100 80 60 40 20 0
ClpB(wt) + ClpB(B1,B2)
Fig. 6. Disaggregation of GFP-38 by mixtures of ClpB(wt) and ClpB mutants in
conjunction with DnaK, DnaJ, and GrpE. (A) The rate of reactivation of
heat-inactivated GFP-38 by ClpB(wt) or ClpB Walker B mutants was measured
with and without DnaK/DnaJ/GrpE as described in the Methods.(B) The rate
of GFP-38 reactivation by mixtures of ClpB(wt) and ClpB(B1,B2) with DnaK/DnaJ/
GrpE was measured as described in the Methods and plotted as a function of
percent ClpB(B1,B2). The dashed gray line represents the linear decrease ex-
pected for a probabilistic mechanism where the activity of the hexamer is
proportional to the number of wild-type subunits. Data are means ⫾SEM (n⫽
3). Some error bars are hidden by plot symbols.
Hoskins et al. PNAS
兩
December 29, 2009
兩
vol. 106
兩
no. 52
兩
22237
BIOCHEMISTRY
described in ref. 38. Protein concentrations given are for monomeric GFP,
GFP-15, GFP-38, MDH, and DnaK; dimeric RepA, DnaJ, and GrpE; hexameric
ClpB; and tetradecameric GroELtrap.
Assays. GFP-fusion protein unfolding (16), GFP reactivation (16), MDH disaggre-
gation (23), RepA activation (16), and ATPase (16) assays were performed as
described with slight modifications described in the SI Methods. GFP-38 reacti-
vation was performed as described in the SI Methods. For subunit mixing exper-
iments, control experiments were conducted to demonstrate that the activity of
ClpB(wt), ClpB(B1), and ClpB(B2) decreased linearly upon dilution over the range of
concentrations used.
ACKNOWLEDGMENTS. We thank Danielle Johnston, Jodi Camberg, Marika
Miot, and Olivier Genest for critical reading of the manuscript and helpful
discussions. We thank Michal Zolkiewski (Kansas State University) for the
plasmid pET20b-K611T and BK Lee (National Cancer Institute) for generating
mathematical models. This work was supported by the Intramural Research
Program of the National Institutes of Health (NIH) National Cancer Institute
Center for Cancer Research.
1. Doyle SM, Wickner S (2009) Hsp104 and ClpB: Protein disaggregating machines. Trends
Biochem Sci 34:40– 48.
2. Mogk A, Haslberger T, Tessarz P, Bukau B (2008) Common and specific mechanisms of
AAA⫹proteins involved in protein quality control. Biochem Soc Trans 36:120–125.
3. Zolkiewski M (2006) A camel passes through the eye of a needle: Protein unfolding
activity of Clp ATPases. Mol Microbiol 61:1094–1100.
4. Parsell DA, Kowal AS, Singer MA, Lindquist S (1994) Protein disaggregation mediated
by heat-shock protein Hsp104. Nature 372:475–478.
5. Mogk A, et al. (1999) Identification of thermolabile Escherichia coli proteins: Preven-
tion and reversion of aggregation by DnaK and ClpB. EMBO J 18:6934– 6949.
6. Bukau B, Horwich AL (1998) The Hsp70 and Hsp60 chaperone machines. Cell 92:351–
366.
7. Lee S, et al. (2003) The structure of ClpB: A molecular chaperone that rescues proteins
from an aggregated state. Cell 115:229–240.
8. Lee S, Choi JM, Tsai FT (2007) Visualizing the ATPase cycle in a protein disaggregating
machine: Structural basis for substrate binding by ClpB. Mol Cell 25:261–271.
9. Wendler P, et al. (2007) Atypical AAA⫹subunit packing creates an expanded cavity for
disaggregation by the protein-remodeling factor Hsp104. Cell 131:1366–1377.
10. Wendler P, et al. (2009) Motor mechanism for protein threading through Hsp104. Mol
Cell 34:81–92.
11. Hanson PI, Whiteheart SW (2005) AAA⫹proteins: Have engine, will work. Nat Rev Mol
Cell Biol 6:519–529.
12. Ogura T, Whiteheart SW, Wilkinson AJ (2004) Conserved arginine residues implicated
in ATP hydrolysis, nucleotide-sensing, and inter-subunit interactions in AAA and AAA⫹
ATPases. J Struct Biol 146:106–112.
13. Sauer RT, et al. (2004) Sculpting the proteome with AAA(⫹) proteases and disassembly
machines. Cell 119:9–18.
14. Tessarz P, Mogk A, Bukau B (2008) Substrate threading through the central pore of the
Hsp104 chaperone as a common mechanism for protein disaggregation and prion
propagation. Mol Microbiol 68:87–97.
15. Weibezahn J, et al. (2004) Thermotolerance requires refolding of aggregated proteins
by substrate translocation through the central pore of ClpB. Cell 119:653–665.
16. Doyle SM, et al. (2007) Asymmetric deceleration of ClpB or Hsp104 ATPase activity
unleashes protein-remodeling activity. Nat Struct Mol Biol 14:114–122.
17. Weber-Ban EU, Reid BG, Miranker AD, Horwich AL (1999) Global unfolding of a
substrate protein by the Hsp100 chaperone ClpA. Nature 401:90–93.
18. Mogk A, et al. (2003) Roles of individual domains and conserved motifs of the AAA⫹
chaperone ClpB in oligomerization, ATP hydrolysis, and chaperone activity. J Biol Chem
278:17615–17624.
19. Weibezahn J, Schlieker C, Bukau B, Mogk A (2003) Characterization of a trap mutant
of the AAA⫹chaperone ClpB. J Biol Chem 278:32608–32617.
20. Werbeck ND, Schlee S, Reinstein J (2008) Coupling and dynamics of subunits in the
hexameric AAA⫹chaperone ClpB. J Mol Biol 378:178–190.
21. Haslberger T, et al. (2008) Protein disaggregation by the AAA⫹chaperone ClpB
involves partial threading of looped polypeptide segments. Nat Struct Mol Biol
15:641–650.
22. Wickner S, Hoskins J, McKenney K (1991) Function of DnaJ and DnaK as chaperones in
origin-specific DNA binding by RepA. Nature 350:165–167.
23. Doyle SM, Hoskins JR, Wickner S (2007) Collaboration between the ClpB AAA⫹
remodeling protein and the DnaK chaperone system. Proc Natl Acad Sci USA
104:11138–11144.
24. Schlieker C, Tews I, Bukau B, Mogk A (2004) Solubilization of aggregated proteins by
ClpB/DnaK relies on the continuous extraction of unfolded polypeptides. FEBS Lett
578:351–356.
25. Mogk A, et al. (2003) Refolding of substrates bound to small Hsps relies on a disag-
gregation reaction mediated most efficiently by ClpB/DnaK. J Biol Chem 278:31033–
31042.
26. Martin A, Baker TA, Sauer RT (2005) Rebuilt AAA⫹motors reveal operating principles
for ATP-fuelled machines. Nature 437:1115–1120.
27. Enemark EJ, Joshua-Tor L (2008) On helicases and other motor proteins. Curr Opin
Struct Biol 18:243–257.
28. Mezard C, Davies AA, Stasiak A, West SC (1997) Biochemical properties of RuvBD113N:
A mutation in helicase motif II of the RuvB hexamer affects DNA binding and ATPase
activities. J Mol Biol 271:704–717.
29. Hishida T, Iwasaki H, Han YW, Ohnishi T, Shinagawa H (2003) Uncoupling of the ATPase
activity from the branch migration activity of RuvAB protein complexes containing
both wild-type and ATPase-defective RuvB proteins. Genes Cells 8:721–730.
30. Moreau MJ, McGeoch AT, Lowe AR, Itzhaki LS, Bell SD (2007) ATPase site architecture
and helicase mechanism of an archaeal MCM. Mol Cell 28:304–314.
31. Crampton DJ, Mukherjee S, Richardson CC (2006) DNA-induced switch from indepen-
dent to sequential dTTP hydrolysis in the bacteriophage T7 DNA helicase. Mol Cell
21:165–174.
32. Singleton MR, Sawaya MR, Ellenberger T, Wigley DB (2000) Crystal structure of T7 gene
4 ring helicase indicates a mechanism for sequential hydrolysis of nucleotides. Cell
101:589– 600.
33. Hoskins JR, Kim SY, Wickner S (2000) Substrate recognition by the ClpA chaperone
component of ClpAP protease. J Biol Chem 275:35361–35367.
34. Hoskins JR, Yanagihara K, Mizuuchi K, Wickner S (2002) ClpAP and ClpXP degrade
proteins with tags located in the interior of the primary sequence. Proc Natl Acad Sci
USA 99:11037–11042.
35. Zolkiewski M, Kessel M, Ginsburg A, Maurizi MR (1999) Nucleotide-dependent oli-
gomerization of ClpB from Escherichia coli.Protein Sci 8:1899–1903.
36. Skowyra D, Wickner S (1993) The interplay of the GrpE heat shock protein and Mg2⫹
in RepA monomerization by DnaJ and DnaK. J Biol Chem 268:25296–25301.
37. Hoskins JR, Wickner S (2006) Two peptide sequences can function cooperatively to
facilitate binding and unfolding by ClpA and degradation by ClpAP. Proc Natl Acad Sci
USA 103:909–914.
38. Hoskins JR, Pak M, Maurizi MR, Wickner S (1998) The role of the ClpA chaperone in
proteolysis by ClpAP. Proc Natl Acad Sci USA 95:12135–12140.
39. Diemand AV, Lupas AN (2006) Modeling AAA⫹ring complexes from monomeric
structures. J Struct Biol 156:230–243.
22238
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0911937106 Hoskins et al.